SavedAdd your journal

Journals / İstatistik Araştırma Dergisi / 2026 / Cilt: 16 Sayı: 1

Adaptive Kalman Filtering for High-Frequency Stock Forecasting

Yüksek Frekanslı Hisse Senedi Tahmini için Adaptif Kalman Filtresi

Levent Özbek(ANKARA ÜNİVERSİTESİ)0000-0003-1018-3114
Pages
1–11
Publication Date
2026-07-31
DOI
—
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

This work contributes to the financial signal processing literature by integrating second-order market dynamics into recursive filtering architectures and highlighting the practical benefits of adaptive noise modeling in high-frequency settings.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Adaptive Kalman filtering techniques are used to demonstrate significant improvements in forecasting accuracy within high-frequency financial time series.
📊
Key Findings & Data:Utilizing real 1-minute interval data from GARAN, a major Turkish banking stock, we adopt a physically inspired three-dimensional state vector comprising price, velocity, and acceleration to capture nuanced market dynamics.
💡
Conclusion & Impact:This work contributes to the financial signal processing literature by integrating second-order market dynamics into recursive filtering architectures and highlighting the practical benefits of adaptive noise modeling in high-frequency settings.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Kalman Filtresi · Hisse Senedi Tahmini · Türk Banka Hissesi

Abstract

Adaptive Kalman filtering techniques are used to demonstrate significant improvements in forecasting accuracy within high-frequency financial time series. This study presents a comprehensive comparative evaluation of four Extended Kalman Filter (EKF) variants—Standard EKF, Adaptive-R EKF, Forgetting Factor EKF, and Mahalanobis Distance EKF—for modeling and forecasting high-frequency stock prices. Utilizing real 1-minute interval data from GARAN, a major Turkish banking stock, we adopt a physically inspired three-dimensional state vector comprising price, velocity, and acceleration to capture nuanced market dynamics. Our methodology systematically incorporates adaptive mechanisms to handle time-varying measurement noise and anomalous observations. The performance of each EKF variant is assessed using four standard metrics: Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Coefficient of Determination (R²). Among the evaluated filters, the Adaptive-R EKF demonstrates superior prediction accuracy while maintaining interpretable and robust state estimates. This work contributes to the financial signal processing literature by integrating second-order market dynamics into recursive filtering architectures and highlighting the practical benefits of adaptive noise modeling in high-frequency settings.

Özet

Adaptif Kalman filtreleme teknikleri, yüksek frekanslı finansal zaman serilerinde tahmin doğruluğunda önemli iyileşmeler sağlamak için kullanılmaktadır. Bu çalışma, yüksek frekanslı hisse senedi fiyatlarının modellenmesi ve tahmini için dört Genişletilmiş Kalman Filtresi (EKF) varyantının—Standart EKF, Adaptive-R EKF, Unutma Faktörü EKF ve Mahalanobis Uzaklığı EKF—kapsamlı karşılaştırmalı bir değerlendirmesini sunmaktadır. GARAN, Türkiye’nin önde gelen banka hisselerinden biri, için gerçek 1 dakikalık aralıklarla elde edilen veriler kullanılarak; fiyat, hız ve ivmeden oluşan, fiziksel temelli üç boyutlu bir durum vektörü benimsenmiş ve böylece piyasa dinamiklerindeki ince farklılıklar yakalanmıştır. Yöntemimiz, zamanla değişen ölçüm gürültüsü ve aykırı gözlemleri ele almak için sistematik biçimde adaptif mekanizmaları içermektedir. Her bir EKF varyantının performansı, dört standart metrik kullanılarak değerlendirilmiştir: Kök Ortalama Kare Hata (RMSE), Ortalama Mutlak Hata (MAE), Ortalama Mutlak Yüzde Hatası (MAPE) ve Belirlilik Katsayısı (R²). Değerlendirilen filtreler arasında, Adaptive-R EKF hem üstün tahmin doğruluğu göstermekte hem de yorumlanabilir ve sağlam durum kestirimlerini korumaktadır. Bu çalışma, finansal sinyal işleme literatürüne ikinci dereceden piyasa dinamiklerini yinelenen filtreleme yapıları içine entegre ederek ve yüksek frekanslı ortamlarda adaptif gürültü modellemenin pratik faydalarını vurgulayarak katkı sağlamaktadır.

🔖

Cite This Article

Özbek, L. (2026). Adaptive Kalman Filtering for High-Frequency Stock Forecasting. İstatistik Araştırma Dergisi, Cilt: 16 Sayı: 1, 1–11. https://ajindex.com/articles/1617745

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (16)

  1. Avellaneda, M., Lee, J. H. (2010). Statistical arbitrage in the US equities market. Quantitative Finance, 10(7), 761–782.
  2. Durbin, J., Koopman, S. J. (2001). Time Series Analysis by State Space Methods. Oxford University Press.
  3. Chen, R., Liu, J. S. (2011). Predictive Filtering for Asset Allocation. Journal of Financial Econometrics, 9(1), 1–28.
  4. Simon, D. (2006). Optimal State Estimation: Kalman, H Infinity, and Nonlinear Approaches. Wiley-Interscience.
  5. Wan, E. A., van der Merwe, R. (2000). The Unscented Kalman Filter for Nonlinear Estimation. In Proc. Of the IEEE Symposium on Adaptive Systems for Signal Processing, Communications, and Control.
  6. Ghosh, A., Dey, S., Chakraborty, S. (2017). Adaptive Extended Kalman Filter in Portfolio Optimization. Applied Soft Computing, 60, 758–770.
  7. Bar-Shalom, Y., Li, X. R., Kirubarajan, T. (2001). Estimation with Applications to Tracking and Navigation. Wiley.
  8. Özbek, L., Özlale, U. (2005). Employing the extended Kalman filter in measuring the output gap. Journal of Economic Dynamics and Control, 29(9), 1611–1622.
  9. Özbek, L., Özlale, U. (2010). Analysis of real oil prices via trend-cycle decomposition. Energy Policy, 38(7), 3676–3683.
  10. Özlale, U., Özbek, L. (2008). Analyzing time-varying effects of potential output growth shocks. Economics Letters, 98(3), 294–300.
  11. Sezgin Alp, O., Özbek, L., Canbaloglu, B. (2023). An analysis of stock market prices by using extended Kalman filter: The US and China cases. Journal of Chinese Economic and Foreign Trade Studies, 16(1), 67–82.
  12. Özbek, L., Hacıoglu, V., Koc，U. (2024). An empirical analysis of consumption and current account in an intertemporal stochastic model. Communications in Statistics - Simulation and Computation, 1–18.
  13. Özbek, L., Özlale, U. (2005). Employing the extended Kalman filter in measuring the output gap. Journal of Economic Dynamics and Control, 29(9), 1611–1622.
  14. Özbek, L., Özlale, U. (2010). Analysis of real oil prices via trend-cycle decomposition. Energy Policy, 38(7), 3676–3683.
  15. Sezgin Alp, O., Özbek, L., Canbaloglu, B. (2023). An analysis of stock market prices by using extended Kalman filter: The US and China cases. Journal of Chinese Economic and Foreign Trade Studies, 16(1), 67–82.
  16. Özbek, L., Hacıoglu, V., Koc，U. (2024). An empirical analysis of consumption and current account in an intertemporal stochastic model. Communications in Statistics - Simulation and Computation, 1–18.